2011 PSLE Discussions and Strategy

[linden2000]

http://www.facebook.com/notes/ministry-of-education-singapore/measures-in-place-in-exam-and-admissions-systems-to-relieve-pressure-on-students/10150302450237004


This note seems to suggest grades are not bell curve-fitted or am I interpreting it wrongly?

[Otah]

ngbrdad:

jtoh:

For languages, a school's standard can be very different from the national standard. Some schools have very high standards for English or Chinese and it would be tough scoring A* in school papers. But compared against the nation, these kids would perform very well. Similarly, it could be that in a school where the standard of English or Chinese paper is average, it might be possible to score an A* consistently in a school paper but not at PSLE.

But is it usual for a kid to score a distinction for higher Chinese but only an A for Chinese ?

Possible when the compo went out of point. Also possible to GEP/ Twining class pupils since from P4 to P6SA1 they only tested on HCL papers and lack of focus on CL paper.

[supermumto3]

Hi 2ppaamm


What you mentioned on your post dated 30 Nov 2011 regarding the transfer credits sounds new to me… Could you shed light on this???

Thanks a mil

[2ppaamm]

linden2000:

http://www.facebook.com/notes/ministry-of-education-singapore/measures-in-place-in-exam-and-admissions-systems-to-relieve-pressure-on-students/10150302450237004


This note seems to suggest grades are not bell curve-fitted or am I interpreting it wrongly?

I think that article came after a lot of talk about PSLE being on the curve and the problems it presents. Specifically it is clear that there is no quota to the number of A* or As in any subject, this means that MOE does not determine the number of A*s and As based on the bell curve. If a student gets >91, he gets an A*. It means A* (or any grade) is based upon the absolute score of the child and not in comparison to the cohort. That's clear.

What's not mentioned is the way the T-score is calculated. For each subject, the T-score = 50+10*[(x-mean)/s.d.]

Compare that to the bell curve or normdist calculation of z-value, z=(x-mean)/s.d.

In our PSLE T-score, they multiple that z value by 10 and then add 50 to it. These are constants and therefore, no matter how you plot it, it will be a bell-curve. Hence, the subject subject T-score is a PERFECT bell curve.

The T-score for the 4 subjects are then added for each child for the PSLE T-score. Hence, while the PSLE T-score may not be a perfect bell curve, it is an aggregation of 4 (or 3) bell curves.

Hope I make sense?

If you follow me so far, it would mean that if the exam paper is easy that year, the mean will be very high, and assuming the s.d. is constant from year to year, your child's T-score will still be around the same, but the number of A* and A will increase. The reverse is true for difficult papers.

Now if you are still with me, if you have a very diverse group of student that year, then the s.d. could be very high, and we hold the difficulty of the paper constant, the range of the PSLE T-score in the cohort is large compared to a cohort with more uniform performers.

Now, we have the PSLE T-score cracked. Hopefully those who have more insights can shed more light. I am looking at it at a pure mathematical/statistical view. Which should apply.

And then a caveat. The PSLE is conducted by MOE only and no external or independent party is involved in its audit. Hence, MOE can manipulate it in anyway it wants/deems fit. It is answerable to nobody on how each child's T-score is calculated. This is a Singapore only exam, and therefore nobody else in the world uses its mechanism, or recognize its purpose. So PSLE is purposeless outside Singapore and even in Singapore, is used as a placement exam only at secondary school level.

So, if the T-score deviates, nobody can understand why or audit the process. It will be a mystery. :scratchhead: Hope this helps!

[2ppaamm]

supermumto3:

Hi 2ppaamm


What you mentioned on your post dated 30 Nov 2011 regarding the transfer credits sounds new to me... Could you shed light on this???

Thanks a mil

Can you clarify which post you mean? Is it Advanced Placement from College Board which NUSH students can opt to take at the end of Year 6?
http://www.collegeboard.com/student/testing/ap/about.html

[jtoh]

2ppaamm:

linden2000:

http://www.facebook.com/notes/ministry-of-education-singapore/measures-in-place-in-exam-and-admissions-systems-to-relieve-pressure-on-students/10150302450237004


This note seems to suggest grades are not bell curve-fitted or am I interpreting it wrongly?

I think that article came after a lot of talk about PSLE being on the curve and the problems it presents. Specifically it is clear that there is no quota to the number of A* or As in any subject, this means that MOE does not determine the number of A*s and As based on the bell curve. If a student gets >91, he gets an A*. It means A* (or any grade) is based upon the absolute score of the child and not in comparison to the cohort. That's clear.

What's not mentioned is the way the T-score is calculated. For each subject, the T-score = 50+10*[(x-mean)/s.d.]

Compare that to the bell curve or normdist calculation of z-value, z=(x-mean)/s.d.

In our PSLE T-score, they multiple that z value by 10 and then add 50 to it. These are constants and therefore, no matter how you plot it, it will be a bell-curve. Hence, the subject subject T-score is a PERFECT bell curve.

The T-score for the 4 subjects are then added for each child for the PSLE T-score. Hence, while the PSLE T-score may not be a perfect bell curve, it is an aggregation of 4 (or 3) bell curves.

Hope I make sense?

If you follow me so far, it would mean that if the exam paper is easy that year, the mean will be very high, and assuming the s.d. is constant from year to year, your child's T-score will still be around the same, but the number of A* and A will increase. The reverse is true for difficult papers.

Now if you are still with me, if you have a very diverse group of student that year, then the s.d. could be very high, and we hold the difficulty of the paper constant, the range of the PSLE T-score in the cohort is large compared to a cohort with more uniform performers.

Now, we have the PSLE T-score cracked. Hopefully those who have more insights can shed more light. I am looking at it at a pure mathematical/statistical view. Which should apply.

Hi Pam,

Thanks for the explanation. I think I understand it.

A question. If as MOE says, the number of A/A* is not pre-determined and that a score of >91 secures you an A* and >70 an A, why is it from year to year, regardless of whether the paper is difficult or easy, the number of A/A* for Math/English/Science usually hovers around 40% and CL usually around 80%? Or is it the case that in a year of difficult Math paper, there are many more As than A* but the statistics do not reveal this as there's no breakdown into A and A*?

[2ppaamm]

jtoh:

2ppaamm:

[quote=\"linden2000\"]http://www.facebook.com/notes/ministry-of-education-singapore/measures-in-place-in-exam-and-admissions-systems-to-relieve-pressure-on-students/10150302450237004


This note seems to suggest grades are not bell curve-fitted or am I interpreting it wrongly?

I think that article came after a lot of talk about PSLE being on the curve and the problems it presents. Specifically it is clear that there is no quota to the number of A* or As in any subject, this means that MOE does not determine the number of A*s and As based on the bell curve. If a student gets >91, he gets an A*. It means A* (or any grade) is based upon the absolute score of the child and not in comparison to the cohort. That's clear.

What's not mentioned is the way the T-score is calculated. For each subject, the T-score = 50+10*[(x-mean)/s.d.]

Compare that to the bell curve or normdist calculation of z-value, z=(x-mean)/s.d.

In our PSLE T-score, they multiple that z value by 10 and then add 50 to it. These are constants and therefore, no matter how you plot it, it will be a bell-curve. Hence, the subject subject T-score is a PERFECT bell curve.

The T-score for the 4 subjects are then added for each child for the PSLE T-score. Hence, while the PSLE T-score may not be a perfect bell curve, it is an aggregation of 4 (or 3) bell curves.

Hope I make sense?

If you follow me so far, it would mean that if the exam paper is easy that year, the mean will be very high, and assuming the s.d. is constant from year to year, your child's T-score will still be around the same, but the number of A* and A will increase. The reverse is true for difficult papers.

Now if you are still with me, if you have a very diverse group of student that year, then the s.d. could be very high, and we hold the difficulty of the paper constant, the range of the PSLE T-score in the cohort is large compared to a cohort with more uniform performers.

Now, we have the PSLE T-score cracked. Hopefully those who have more insights can shed more light. I am looking at it at a pure mathematical/statistical view. Which should apply.

Hi Pam,

Thanks for the explanation. I think I understand it.

A question. If as MOE says, the number of A/A* is not pre-determined and that a score of >91 secures you an A* and >70 an A, why is it from year to year, regardless of whether the paper is difficult or easy, the number of A/A* for Math/English/Science usually hovers around 40% and CL usually around 80%? Or is it the case that in a year of difficult Math paper, there are many more As than A* but the statistics do not reveal this as there's no breakdown into A and A*?[/quote]Not sure about that 40% and 80% thingy. I added a caveat.

[jtoh]

No problem Pam.

[linden2000]

2ppaamm:

linden2000:

http://www.facebook.com/notes/ministry-of-education-singapore/measures-in-place-in-exam-and-admissions-systems-to-relieve-pressure-on-students/10150302450237004


This note seems to suggest grades are not bell curve-fitted or am I interpreting it wrongly?

I think that article came after a lot of talk about PSLE being on the curve and the problems it presents. Specifically it is clear that there is no quota to the number of A* or As in any subject, this means that MOE does not determine the number of A*s and As based on the bell curve. If a student gets >91, he gets an A*. It means A* (or any grade) is based upon the absolute score of the child and not in comparison to the cohort. That's clear.

What's not mentioned is the way the T-score is calculated. For each subject, the T-score = 50+10*[(x-mean)/s.d.]

Compare that to the bell curve or normdist calculation of z-value, z=(x-mean)/s.d.

In our PSLE T-score, they multiple that z value by 10 and then add 50 to it. These are constants and therefore, no matter how you plot it, it will be a bell-curve. Hence, the subject subject T-score is a PERFECT bell curve.

The T-score for the 4 subjects are then added for each child for the PSLE T-score. Hence, while the PSLE T-score may not be a perfect bell curve, it is an aggregation of 4 (or 3) bell curves.

Hope I make sense?

If you follow me so far, it would mean that if the exam paper is easy that year, the mean will be very high, and assuming the s.d. is constant from year to year, your child's T-score will still be around the same, but the number of A* and A will increase. The reverse is true for difficult papers.

Now if you are still with me, if you have a very diverse group of student that year, then the s.d. could be very high, and we hold the difficulty of the paper constant, the range of the PSLE T-score in the cohort is large compared to a cohort with more uniform performers.

Now, we have the PSLE T-score cracked. Hopefully those who have more insights can shed more light. I am looking at it at a pure mathematical/statistical view. Which should apply.

And then a caveat. The PSLE is conducted by MOE only and no external or independent party is involved in its audit. Hence, MOE can manipulate it in anyway it wants/deems fit. It is answerable to nobody on how each child's T-score is calculated. This is a Singapore only exam, and therefore nobody else in the world uses its mechanism, or recognize its purpose. So PSLE is purposeless outside Singapore and even in Singapore, is used as a placement exam only at secondary school level.

So, if the T-score deviates, nobody can understand why or audit the process. It will be a mystery. :scratchhead: Hope this helps!

Thanks for the clarification.
Generally easy paper means higher number of A and A* since these are not bell-curve fitted. But not necessarily higher t-scores because of higher mean (since T-scores are calculated relative to the cohort).

[2ppaamm]

linden2000:

Thanks for the clarification.
Generally easy paper means higher number of A and A* since these are not bell-curve fitted. But not necessarily higher t-scores because of higher mean (since T-scores are calculated relative to the cohort).

Yes, that's about it.